Sparse Data Analysis Strategy for Neural Spike Classification

Table 5

Tables of cross-validation of the data benchmarks.

(a)

Fisher Iris								Township
Known classes	Clusters PB-Clus			Known classes	Clusters k-means			Known classes	Clusters PB-Clus
	1	2	3		1	2	3		1	2	3	4
Setosa	50	0	0	Setosa	50	0	0	Urban cities	8	0	0	0
Versicolor	0	50	0	Versicolor	0	49	1	Transitions	0	4	0	0
Virginica	0	17	33	Virginica	0	13	37	Country towns	0	0	2	0
								Unclassified	0	1	0	1
Classification rate =0.88				Classification rate = 0.90				Classification rate = 0.94

(b)

Ruspini					Faithful			Geysers
Clusters k-means	Clusters PB-Clus				Clusters k-means	Clusters PB- Clus		Clusters k-means	Clusters PB-Clus
	1	2	3	4		1	2		1	2	3
Group 1	50	0	0	0	Group 1	168	4	Group 1	88	2	7
Group 2	0	35	0	0	Group 2	0	100	Group 2	0	105	0
Group 3	0	0	15	0				Group 3	0	0	97
Group 4	0	0	0	20
Classification rate = 1.00					Classification rate = 0.98			Classification rate = 0.97